US 12,406,479 B2
Method and system for dynamically generating annotated content for AR-VR applications
Gopichand Agnihotram, Bangalore (IN); and Shrivardhan Satish Suryawanshi, Sangli (IN)
Assigned to Wipro Limited, Bangalore (IN)
Filed by Wipro Limited, Bangalore (IN)
Filed on Mar. 31, 2022, as Appl. No. 17/657,436.
Claims priority of application No. 202241007054 (IN), filed on Feb. 10, 2022.
Prior Publication US 2023/0252768 A1, Aug. 10, 2023
Int. Cl. G06V 10/774 (2022.01); G06V 10/25 (2022.01); G06V 10/40 (2022.01); G06V 10/74 (2022.01)
CPC G06V 10/7747 (2022.01) [G06V 10/25 (2022.01); G06V 10/40 (2022.01); G06V 10/761 (2022.01)] 14 Claims
OG exemplary drawing
 
1. A method of dynamically generating annotated content for training a model for Augmented and Virtual Reality (AR-VR) applications, the method comprising:
receiving, by a processor of a content annotating system, a plurality of images for an object associated with AR-VR applications;
obtaining, by the processor of the content annotating system, one or more pre-annotated datasets related to one or more images of the plurality of images from a user;
generating, by the processor of the content annotating system, a plurality of augmented image datasets for the plurality of images using a predefined set of augmented parameters, wherein the predefined set of augmented parameters comprise distortion associated with each of the plurality of images;
extracting, by the processor of the content annotating system, a set of features from the one or more pre-annotated datasets and the plurality of augmented image datasets, wherein the extracted set of features indicate a vector component comprising textual features, color features, and edge features, wherein the extraction comprises:
selecting, by the processor of the content annotating system, one or more Region of Interests (ROIs) from each of the one or more pre-annotated datasets; and
extracting, by the processor of the content annotating system, the set of features from the one or more ROIs associated with each of the one or more pre-annotated datasets and the set of features for the plurality of augmented image datasets, wherein the extracted set of features of the one or more pre-annotated datasets is compared with the set of features of the plurality of augmented image datasets, wherein the comparison comprises:
computing, by the processor of the content annotating system, a similarity value between the set of features of the one or more pre-annotated datasets and the set of features of the plurality of augmented image datasets; and
comparing, by the processor of the content annotating system, the similarity value with a threshold similarity value to identify the one or more ROIs for each of the plurality of augmented image datasets; and
generating, by the processor of the content annotating system, annotated content for the plurality of augmented image datasets based on the comparison, wherein the annotated content for the plurality of augmented image datasets along with the one or more pre-annotated datasets are used for training the model associated with the AR-VR applications.